Scales and design - Page 4

Want to talk espresso but not sure which forum? If so, this is the right one.
wojtowip

#31: Post by wojtowip »

I saw the response graphs and they do look like a curve, so will need to find the right formula to convert resistance to weight, I think there is one given, but I'll need to double check.

Looking at the specs, it looks like the force sensors are pretty consistent per unit, but can vary a lot more between devices, so you would need to calibrate each one individually.

With 2% repeatability, seems like 30g should only vary about 0.6g, so probably decent enough for pulling shots, about as good as I can get anyway stopping. Might not be .1g precise, but for a shot getting within a gram is pretty good for me.

Will have to see how these actually hold up once I get them. Otherwise the load cell looks pretty promising. I could save a lot of space by just drilling a hole in the bottom plate and mounting it directly to the base, that would save the thickness of the bottom plate at least. Those sensors should be a lot more stable.

Never expected to be dusting off my Arduino skills when making coffee :).
LMWDP #694

jpender

#32: Post by jpender »

wojtowip wrote:I saw the response graphs and they do look like a curve, so will need to find the right formula to convert resistance to weight, I think there is one given, but I'll need to double check.
All I saw was a graph. Probably you'll have to take a bunch of measurements, plot weight vs resistance (or voltage or A/D count), and then use something like Excel to generate a best fit polynomial.

wojtowip wrote:Looking at the specs, it looks like the force sensors are pretty consistent per unit, but can vary a lot more between devices, so you would need to calibrate each one individually.

With 2% repeatability, seems like 30g should only vary about 0.6g, so probably decent enough for pulling shots, about as good as I can get anyway stopping. Might not be .1g precise, but for a shot getting within a gram is pretty good for me.
Unless you're using a styrofoam cup the weight will actually be quite a bit higher. How will the device respond to small incremental weight changes? That's not obvious to me. Let us know how it works out for you.

Flair Espresso: handcrafted espresso. cafe-quality shots, anytime, anywhere
Sponsored by Flair Espresso
Espressoman007 (original poster)

#33: Post by Espressoman007 (original poster) »

wojtowip wrote: With 2% repeatability, seems like 30g should only vary about 0.6g, so probably decent enough for pulling shots, about as good as I can get anyway stopping. Might not be .1g precise, but for a shot getting within a gram is pretty good for me.

Never expected to be dusting off my Arduino skills when making coffee :).
Motivation is the biggest driving force!
I was thinking that if its accuracy is within 2g, it would be be more than OK for me.
I am not splitting hairs with seconds and grams, as long as the espresso is tasty...and with Strietman and a nice grinder that range is waaay bigger than 2sec or 2g.

You'll make it!

Cheers!

wojtowip

#34: Post by wojtowip »

Yeah, not sure how it will respond to incremental weight, that's kind of why I want to play with it. If I remember correctly the base was about 500g and a cup ends up being another 100-200g, so looking at a total change from about 600g to 630g. I'm really curious how it will work. Would be nice if it was close. This would make a very low profile sensor. The fallback is a load cell, but it's hard to get a hand on low profile ones.

Now thinking about I'm having doubts the force sensor will work well, unless incremental changes work correctly. 2% of 600g is 12g. So if I could zero out on 600g and incremental error would be 2% of the 30g change that would make it usable. if the difference would be 12g, that would be a lot less usable.
LMWDP #694

wojtowip

#35: Post by wojtowip »

Alright, so I finally got a chance to play around with the force sensor and a multimeter (since it's just a resistor) and it's a no go. It's more of a instantaneous reading device to get how hard something was pushed for a moment and then meant to let go. It's not designed to weigh an object for an extended period of time.

So if I put a 100g weight on the pad, it will register around 9kohm initially, but that resistance will keep dropping slowly till it was below 5k a minute later and still dropping. So looks like this is good to just measure at an instant, but not so much for something like pouring coffee out, since it doesn't provide enough resistance back to the weight being applied.

Bummer. Oh well onto the load cells next. The Amazon platform I ordered is huge, so will need to modify it a bit to make it work inside, but first want to check how well I can weigh on it.

Also got a quote back from SMD about their lower profile high accuracy load cells and they are about $199, so kind of understand why the Acaia scales are so expensive now.
LMWDP #694

randytsuch

#36: Post by randytsuch »

Deleted, was off topic

Randy

jpender

#37: Post by jpender »

wojtowip wrote:So if I put a 100g weight on the pad, it will register around 9kohm initially, but that resistance will keep dropping slowly till it was below 5k a minute later and still dropping. So looks like this is good to just measure at an instant, but not so much for something like pouring coffee out, since it doesn't provide enough resistance back to the weight being applied.
I'd read that these devices aren't suitable for quantitative measurement but that's a lot worse than I expected. I watched a video where someone was placing calibration weights on one and it sure didn't look like the output was dropping fast enough to lose almost half its value in 60 seconds. Did your ohmmeter output look like this or was it dropping right off the bat?

wojtowip wrote:Also got a quote back from SMD about their lower profile high accuracy load cells and they are about $199, so kind of understand why the Acaia scales are so expensive now.
The Pearl and Pyxis have heights of 15mm and 13mm, respectively. These heights are achievable with the less expensive load cells. As I mentioned earlier I'm pretty sure I could hack my inexpensive scale (16mm high) down to an 11-12mm height. I suspect that the Acaia scales are expensive because the load cells they use are both fast and precise, they connect to your phone, and they are a niche product.

Aida Battle: Indigo Reserve from world renowned Finca Kilimanjaro in El Salvador
Sponsored by Aida Battle
jpender

#38: Post by jpender »

randytsuch wrote:I ended up buying a bluetooth scale from Amazon, and was able to figure out how to read the bluetooth data to get the weight.
I'm not sure I understand. What does bluetooth have to do with the scale height? Why not just read the built in display?

The scale you used is specified to be 15mm high, still "too thick" for the OP. It's also a lot wider, about 1cm too wide to fit under my espresso machine.

wojtowip

#39: Post by wojtowip »

jpender wrote:I'd read that these devices aren't suitable for quantitative measurement but that's a lot worse than I expected. I watched video where someone was placing calibration weights on one and it sure didn't look like the output was dropping fast enough to lose almost half its value in 60 seconds. Did your ohmmeter output look like this or was it dropping right off the bat?
I was not getting this stable of a ohm reading as he is in the video. I don't have it hooked up to an Arduino yet, so wasn't able to graph, but the resistance pretty much started dropping as soon as I placed an item on the force pad. I'll try again to see if adding a platform on top to cover the entire surface would make a difference. Maybe I just got a defective part or user error. Could really be either in this case.
LMWDP #694

randytsuch

#40: Post by randytsuch »

jpender wrote:I'm not sure I understand. What does bluetooth have to do with the scale height? Why not just read the built in display?

The scale you used is specified to be 15mm high, still "too thick" for the OP. It's also a lot wider, about 1cm too wide to fit under my espresso machine.
You're right, my bad. Deleted post